Draft, 7 September 1998

9, 11 January 1999

The Nature and Structure of Emotions

by Randall R. Dipert

U.S. Military Academy

West Point NY 10996

e-mail: cr2559@usma.edu

Philosophers have almost always said something about emotions and passions whenever they have discussed human mental life. Many have asserted that it is some emotions or, more broadly, passions, that are to be primarily valued and sought. These valued passionate states of mind might include emotions, moods, desires, belief-like feelings of conviction and commitment, and romantic or erotic love, which are typically scarcely distinguished. Not only are these states of mind lumped together, but the reasons why they are valued may likewise be various: they may be valued because of their intrinsic feeling (especially insofar as they are intense), through their long-term or deep effects on the rest of our practical and mental lives, through their effects on others’ lives, or even in the glimpse they give us of an object that transcends our mundane and superficial concerns, as in love, peak experiences, or intimations of God, Beauty, or Nature. Others have claimed that it is in the subduing or elimination of some or all of these passions that the ideal human life consists. Again, what precisely are the objectionable passions is typically not delineated, and why such mental states are objectionable may be diverse and even unspecified. One might resent their "disruptive" nature on our mental life, especially insofar as some of them stem from external, uncontrollable sources, and instead seek a calm state that is within one’s control and not subject to these whimsical externalities. Or one can see many or all passions as disruptive of control and success in our inner or outer life, or in the lives of others. We might call this latter group the anti-emotional Rationalists, and the former group the pro-emotional Romantics.

Almost all philosophers who have said much about moral psychology and what might be called ethics for states of mind, such as Plato, Aristotle, the Stoics, Hume, and extending into recent contemporay authors such as Nussbaum, can be categorized in this way. Although most major literature articulates nuanced and mixed views about the subject, it is usually difficult to discern the exact nature of the valued or objectionable passionate states of mind, and likewise difficult to see precisely why such states are valued or condemned: the consequentialist reasoning is usually highly speculative, while the "internalist" rhetoric that values certain states of mind for their own sake seems faintly attractive but difficult to talk about very usefully. Furthermore, a certain odd mythology seems to surround the entire discussion, namely that there exist non-passionate states of mind, such as "rational belief," and perhaps that we should seek them--or maybe shun them. In the absence of widely accepted accounts of what states of mind are, and how emotion, desires, beliefs, and other states of mind differ from each other (or are similar), such normative discussions--as well as analytical proposals "reducing" of one state to another--are likely to be more perplexing than illuminating.

In this essay, I would like to propose a definition of the state of mind known as an emotion. Such a definition would ideally characterize the general category to which emotions belong, namely states of mind. More specifically, emotions are intentional states of mind, although this is itself arguable: they seem to be acts or attitudes with an "intended" conceptual target or object. I am angry at Ralph, but admire Elisabeth. Such a definition would also carefully distinguish emotions from other states of mind, such as beliefs, desires, and moods, many of which are similarly intentional. Any good philosophical definition of emotion should provide the beginning of a theory of emotion. A theory of emotion would begin to answer collateral questions about emotions, such as the function they generally serve, their "purpose" in human life, as well as to provide a basis for addressing normative questions, such as whether emotions are generally good or bad for individual human and social life, and if sometimes good, exactly which emotions are good, and why. The main goal of such a speculative theory of the emotions would be, however, to provide a complete and integrated account of all of human mental life: to relate emotions (and their components, such as their intentional objects) to other states of mind. Only then would we begin to form a comprehensive, integrated and thus complete picture of human mental life. The nature of "emotions"--whatever exactly they are--cannot be easily isolated from a number of our other states of mind.

Emotions have not escaped the notice of contemporary philosophy of mind and philosophical psychology, although they occupy a very small role in the whole of modern philosophy. (See (Gardiner et al 1937).) Several non-exclusive strategies or methodologies have arisen in the last decades for the analysis of emotions: intentionalism, cognitivism, functionalism, and behaviorism. Intentionalism claims that emotions are necessarily directed toward "conceptual" objects. Cognitivism sometimes includes the intentionalist claim but lays emphasis on the cognitive, or thought-like, nature of the object, and especially on certain inference-like, even "rational," processes that produce the emotion. (Computationalism, at least in its strong form, often combines an intentionalist view that thoughts are just represented data and the process-emphases of cognitivism in arguing that computations are these processes. See (Deigh 1994).) Functionalism is a widespread but perhaps ultimately ill-defined approach that seeks to analyze a state of mind (and less often mental contents as well) strictly in terms of the role--usually understood as causal role--that a state of mind such as emotion plays in all of mental life. What does or can cause such a mental state to occur, and what other states of mind and behavior does it in turn cause. Finally, behaviorism is a kind of limiting case of functionalism: it treats emotions (or another mental state) strictly in its capacity to produce observable behavior (or more broadly, public symptoms); while Gilbert Ryle is perhaps the best known example, we see the view reemerging in recent attempts to "naturalize" emotion (Griffiths 1997). There are also other approaches with links to contemporary science: there is psychobiology--akin to sociobiology, perhaps better: "evolutionary psychology"--in which we use arguments about mammalian or human evolution to analyze the nature or purpose of emotions. There is also recent research in which we apply very recent findings in psychopharmacology or neurophysiology to speculate about the nature of emotions. In most discussions, it is difficult to determine whether these strategies are employed to define emotion, to define a particular kind of emotion (such as fear), to identify and differentiate but not define them, to individuate emotions, or some other unspecified philosophical or scientific task. Simply aiming to "understand" emotions--such as by appreciating the possible mechanisms in evolution by which they arose-- seems to cover too much ground, and any understanding of emotions per se likely presumes their identification as a well-defined species of mental activity.

In this essay, I propose a view I call the phenomeno-structuralist account of states of mind, such as emotion. Inspired in part by concerns that also motivated David Chalmers’ fine The Conscious Mind, I try to take first-person consciousness seriously. This is the "phenomenal" component, as opposed to what Chalmers--perhaps misleadingly--calls the "psychological" (behavioral, external, empirical) aspect. Second, I collapse cognitivism, functionalism, and intentionalism into one central account: the view that it is not causal roles borrowed from the ill-understood and intuitively-grasped physical realm that constitute our best or ideal "picture" of mental inter-relationships, but purer notions of formal structure that may be unique to the mental world. This is my structuralism. All of the connection of emotion to concepts (intentionalism, cognitivism), the status of mental content itself, and the inter-relationships among mental states (functionalism) are viewed along the same lines, namely, as descriptions of structures of atoms of consciousness I call feelings.

Some of the more obvious weaknesses of contemporary discussions in the philosophy of mind are its preoccupation with certain foundational questions, such as the mind-body problem, the nature of consciousness, or broad investigations of intentionality and its reducibility. In this way, the philosophy of mind has often sidestepped the full richness of human mental life. And it has often sought neo-positivistically to analyze belief in terms of observable behavior, and sometimes even to reduce beliefs to this behavior or to mysterious but sanitary "dispositions" toward behavior. Even where it has concerned itself with substantive aspects of mind, interest has remained preoccupied with the mind’s largely passive epistemic subsystems, notably perception, and especially, belief. That is, contemporary philosophy of mind has often treated desires, emotions and moods--even intentions--as marginal and unusual, even troubled, aspects of human mental life.

While some philosophy attention was called to emotions and emotion-like states earlier in the century by figures such as Heidegger and Sartre, the last decade has seen a wave of interest in emotions especially, but also in kindred passions like desires and moods. Works of authors such as Robert Solomon, Robert Gordon, Ronald de Sousa, Martha Nussbaum,, popular authors such as Anthony Dimasia, Jeffrey Masson, and others come to mind.

There is little doubt, I believe, judged either from our own interior experience, or from the perspective of human history and our best guesses about the workings of others’ minds, that emotions--whatever exactly they are--are an extremely important part of the life of the human mind. They seem to be intrinsically valued, which we can discern in the pleasure we take in emotions and the way we sometimes seek just them. We seem to seek and enjoy even "negative" emotions, such as fear, anger, and pathos, at least in fictional settings. Viewed instrumentally, emotions are both an impelling force in, and the strongest impediment to, life’s grandest projects. Nothing but an emotion, it seems, can ever cause otherwise decent people knowingly to take another person’s life unjustifiably. And likewise, nothing but the sustaining energy that long-term emotions sometimes provide produces noble life work, including masterpieces of art or philosophy. These sources of nobility include such emotions as an abiding sense of injustice, our own pride, envy, and an annoyance at ignorance.

Whether emotions are a necessary part of any mind’s activity, or every rational mind’s activity, is a subsidiary question of considerable interest. My own brief answer to this latter question is that some "valuing" or "motivating" subsystems are a necessary part of agency, and thus necessary for an active mind. Emotions are one such motivating subsystem. The importance of seeing emotions as mental states that attach value, and thus are, in my terminology, motivating, is one of the strongest themes of Robert Solomon’s The Passions. Furthermore, there are reasons to suspect that a full notion of belief, and thus even the passive activities of the mind on which modern philosophy of mind has harped, require a kind of "inner" motivation, namely for spawning inquiry and for guiding inference, of the sort that emotions provide. I will return to these points--which hint at my broader theme of the need for an integrated, comprehensive account of states of mind as an account of any "mind" worthy of the name.

The emotions, and their kindred "marginal" states of mind, such as desire--together often termed "passions"--were of course far from marginal aspects of ancient theories of mind. They are there, in Plato, Aristotle, and the Stoics, strangely to us on a par with rational belief. Plato and Aristotle seem to suggest that emotions, even appetites, are not wholly bad, and that their presence and effects are not necessarily to be minimized. Modern philosophy has of course been largely silent on the topic, except in some minor accounts of moral psychology or the sentiments. The standard textbook account of the patron saint of Modern Philosophy, Descartes, would seem to leave no place in the ideal life of the mind for such riffraff as emotions or desires, fixated as this life should be on "certain" or justified beliefs. (Descartes’ actual view of the emotions and ethics is quite a bit more complex, and thus suggests that this crude historical picture of the overwhelming importance of one kind of rationality--rationality in judgment--is in part rhetoric and in part oversight due no doubt to a reigning standardized textbook/comic book approaches to these masterpieces. )

I presuppose at least the usefulness of intentionalalist conceptions of (most) states of mind. This involves the assumption that there are distinctions between the act or attitude of a state of mind, and its content or target (utlimately both to be analyzed in the full phenomeno-structural account of mind). This intentionalist conception also involves a narrower claim about the nature of this relationship, that it is a targeting or "directing" of thinking toward these thought-objects. This view is originally associated with the names of the 19th century philosophers Brentano and Meinong. In fact, I will argue that states of mind acquire some of their distinctness through their multiple relationships to various kinds of structures (intentional objects), rather than having one single target or content. While I believe that the phenomenal and linguistic naturalness and utility of this notion starts to establish its justifiability, there is other evidence for it. It seems to me that I can simultaneously hold different attitudes to the same thought content, and that I can have the same attitude to different thought contents. For example, I can both believe and be distressed at the fact that a great many people are foolish and wicked. It is perhaps even easier to see that I can have the same attitude to two quite different thoughts. I can believe both that snow is white and, in the same sense of "believe," that grass is green.

Finally, I will increasingly appeal to notions of structure that are not to be equated with the causal relationships proposed by functionalism, nor even with usual unexamined, primitive notion of an intentional relationship. Nor do I suppose, or mean to suggest, that such structure is precisely amenable to physical analysis in terms of forces and space-time (or to whatever one’s physics is), or to a logical or quasi-logical theory, such as does logical atomism or set-theoretic metaphysics. My views about structure--pure and neutral mathematical structure--are described at length elsewhere (Dipert 1997), and it is these and only these to which I ultimately mean to refer. Although I will sometimes mention ordinary concepts and ideas (e.g., of grass), or complete thoughts, as if these everyday facts were perfectly well understood in the manner that contemporary analysis typically presumes, I ultimately hold that the "content" of any concept is purely structural. It is a matter both of its internal structure in the sense of its composition, as well as in its "external" structure in the form of relationships with other concepts.

Before I attempt to state a characterization of emotion, I believe it is extremely helpful to offer an analysis of one paradigmatic emotion, namely anger. It is admitted that my account of anger masquerades as a common-sense, pre-theoretical account of an emotion, but that it in fact applies the machinery of my full, and quite theoretical, account of emotions and indeed any mental state. Some aspects of this account will be familiar to anyone familiar with Aristotle’s views, or with modern cognitive ones, such as Robert Gordon’s. My analysis of S’s anger at P is that it is a state of mind described by this diagram:

By an arrow, ® , I indicate some form of asymmetric, transitive mental linkage or production. By a double arrow, Þ , I indicate some asymmetric, non-transitive "targeting" of a feeling--or what is often called an "attitude"--toward an object. That is, anger for me is intentional in the usual sense of being "directed toward" an object, its intentional object. I will shortly say much more about my technical notion of "feelings," but for now it is sufficient to understand a feeling as the most basic, and simple, event of consciousness, which serves both as the distinctive core and as the binding element of all phenomenological acts.

Views similar to the one I propose are fairly common in the literature, especially in recent proposals that are sympathetic cognitivism, intentionalism, and functionalism. A particularly fine-grained little-appreciated functionalist--indeed, mechanistic--analysis of emotion can be found in (Sloman 1987); another similarly "structural" account is associated with the impressive work of Nico Frijda, such as (Frijda et al. 1995). How my account differs from these now rather commonplace ones will only become fully evident toward the end of this essay: roughly, it is in a neutral account of relationships that distinguish it from the causal ones of functionalism, and still more radically in fusing this "functional" structure with content--and consciousness--itself.

For reasons that will shortly become clearer, I do not think that anger is to be identified as any one component in this diagram. I do not believe, for example, that it is correct to point toward the "feeling" and say that this is the anger. The feeling alone does not include the distinctive intentionality of anger, or its precise object, and it does not include the means by which this feeling was produced, namely the normative beliefs about wrong-doing, that are constitutive of this kind of emotional state I am calling anger. It would be more accurate to say that the anger is the feeling as it is embedded in a certain pattern or relationships to other feelings. My picture suggests that the concept of anger is an "historical" or genetic concept, distinctive because of its past or how it was brought about--distinctive in the possibly uncognized external relationships it has to other states of mind.

Anger is best described as a hostile feeling that is intentional in a certain way, produced in a certain way (namely by certain beliefs or judgments), and has, at least paradigmatically, certain effects in terms of actions and behavior (typically produced by interacting with other states of minds, such as goals, beliefs, and believed social constraints). This account has certain affinities with what are called "functionalist" accounts of states of mind. Anger, or any emotion, is to be identified by its place in a system of mental activities (such as emotion-forming, belief-forming, and behavior-inducing subsystems) and mental states. True anger does not then occur in an entity that lacks all or many of these other subsystems (or "modules," as they are called in cognitive science), these subsystems are themselves constituted partly by their relations to other subsystems, including emotions, and these subsystems must be related to each other in certain ways--such as in having certain beliefs "cause" certain emotions.

It so happens that the English term "anger" is crude and broad, and that I have chosen to focus on characterizing a species of anger as its paradigm that might better be termed indignation. To an extent, then, I will follow ordinary English usage and to an extent I will seek to reform or deepen it. My analysis of what I term anger will eventually allow us to discern various differences among the attitudes commonly called "anger.". These include a still more self-aware form of anger, which we can term "reflective indignation," in which one simultaneously and integratedly knows or has beliefs about the cause of one’s feeling. In what I call "rage," one has a feeling caused by a judgment about another person, but this judgment may be about their behavior (rather than about an intentional action) or by a judgment about their interference with one’s goals or pleasure (rather than about the believed moral wrongness of their intentional action.). We might call these two forms of anger, respectively "behavioral" rage and non-moral rage. I here deviate again from English, since I believe "rage" in its normal usage also suggests a strong emotion and not just a relatively inarticulate one. In my notion of rage, we ignore this quantitative dimension.

A clear contrast with the emotional mental state of anger is offered by what I will call the pure reactive mental state of a "startle" (or "flight") response. (Some have called these primitive emotions.) A loud noise or the sudden appearance of a large object will certainly cause in one a feeling, and perhaps accompanying physiological symptoms and behaviors. Such a reactive state nevertheless differs from paradigm emotions in two ways. First, the precipitating cause of the feeling will necessarily be largely or completely sensory, rather than conceptual as in the case of beliefs about an action proper or about the morality of types of action. The feeling associated with a reactive state will initially lack any attachment to an intentional object. You will not yet have digested, and become aware of, exactly what has startled you. A moment’s reflection on the putative state of affairs causing the startle response may then produce an true emotion of fear, in which judgments of real danger and targeting an intentional object are "filled in." (It will also be the case that the startle response as a reactive state of mind will at most induce more immediate physiological and behavioral responses, and not actions with deliberation and planning to maximize effectiveness.) In the immediate startle response, I believe we can discern an almost pure "feeling." Angst by contrast is a non-primitive, long-term mood, in which the cause and target of the feeling have eluded reliable identification or even conceptualization. It may have physiological and behavioral consequences, but will not extensively and effectively direct action. Both in reactive states and in moods like angst, we see a natural tendency toward, and the practical desirability of their becoming, emotions in my sense. True emotions are, in this sense, rational and desirable, especially compared with moods and mere reactive emotions. Without the identification of target, it is impossible to focus corrective or sustaining action. Reactive emotions, moods, and other unproductive or even counterproductive states of mind may nevertheless retain a raw phenomenal value as feelings of some sort.

There is another related emotional state that is also not produced in the way that indignant anger is, because it is not directed toward an intentional object that is fully conceptualized as a person. We might term this non-person directed rage. Another possibly "defective" form of quasi-anger is the presence of a feeling that is directed toward the wrong person, toward a person that is conceptualized differently from the person judged to cause the feeling. One "takes out" one’s anger caused by school or work on the wrong person or object, such as a family member or one’s dog. In my model, anger is an attitude, but not a propositional attitude. It is directed toward a conceptual object, necessarily an object conceptualized as a person in its proper form. "Orthotic"--right-- indignant anger is properly directed toward persons, and in particular toward the person that is part of one’s conception of the emotive feeling’s cause, because only persons can be expected to recognize, and react properly to this anger. Other entities that are the alleged object of emotions could not recognize behaviors as caused by someone else’s judgment that they have done something morally wrong, and hence be stimulated to reflect on whether it is morally wrong. Rage is a different matter and can be properly directed at any entity that is "reformable," such as a pet or young child, whose behavior I can modify by expressing such rage, without irrationally intending them to recognize the moral wrongness of what I saw in their behavior. Here I may intend to reform their behavior, but not by getting them to reflect first on the moral wrongness of their deed.

There is a related state of mind that we might call annoyance that is directed toward a proposition or believed state of affairs. I believe it is quite obvious that I can be annoyed, and momentarily emotionally engaged, when someone accidentally steps on my toes. I am annoyed at some person’s stepping, or some object’s pressing, on my toes. Likewise, I can be annoyed at the weather and a variety of states that befall me and the world. Non-indignant, non-person-directed rage is appropriate when the intentional object’s "character" is reformable. Annoyance is appropriate when a state of affairs is alterable or sustainable through foreseeable consequences of one’s annoyed feeling. I am thus suggesting that negative, anger-like emotions toward something that is neither conceptualized as a person (for anger) nor as a responsive and hence reformable entity (for rage) is possible only toward a state of affairs, not an object. One cannot be annoyed at the toes in a way I can be angry at a person. "Toes stepping on mine" is a situation that might indeed be avoided in the future by some means or other, but not by having the toes notice my moral indignation, nor by punishing and reforming them. It requires other plans, not connected to the lacking moral consciousness or perceptual responsiveness of the toes. (This implies that annoyance at the weather is non-orthotic and, with deeper reflection, raises Stoic conundrums concerning whether any events or characters are alterable--determinism--and thus whether any such anger-like states are ever appropriate and right.)

It is not necessary for annoyance to have a moral component, for it to be directed at a person. Likewise it is not necessary to believe that the causing entity’s behavior was intentional. But an emotionally well-put-together individual does not--cannot in my model--get angry at the child that stumbled into me when pushed by another child, or at the weather. At least not in this very narrow sense of anger as indignation. Interestingly, one can also be angry at someone without having first been annoyed--that is, having one’s own happiness or pleasure interfered with--although this is a rarefied emotion characteristic only of morally developed people. One can be angry, for example, that a person did something morally wrong to another individual for whose misfortune we directly feel little or nothing. This is in fact the usual use of the term "indignation," but which I would call oblique indignation in contrast to the direct indignation in which the act is also believed to interfere with one’s own interests and happiness.

The consequences or effects of emotions fall into four broad categories. First, there can be immediate and generally unsuppressable physiological consequences of an emotion--a flushed face, sweating, increased heart rate, and so on. These will generally be manifested in proportion to the strength of the feeling. They may be highly variable across individuals, and physiological symptoms associated with anger or rage might be identical to those of quite different emotions or emotional reactions, such as fear. Second, there can be immediate and reflexive behavioral consequences: in adults of normal constitution, these are very rare, but in youth and with incontinent individuals, these might be immediate and close to unsuppressable. These might be an impulse to strike an individual. However, an angry or impatient facial expression or tone of voice also fall in this category (although these are in most individuals reformable and simulatable).

Third, there are considered or deliberated behavioral consequences: these will be actions that result from interaction with other mental states. Such an action might be a decision to harm punish the individual, either immediately or later. Such a decision will result from considerations of beliefs concerning risk, social, legal and ethical ramifications, effectiveness, and means-ends reasoning. It may also involve interactions with the other emotional and conative states, such as liking the person, fearing them, and so on. Such effects may result from habitual processes as much or more as they do from individual instances of conscious deliberation. Finally, there may be an impact the emotion has on other mental states or mechanisms, such as coming to believe that "such people" are to be avoided or feared, or in emotions toward other entities (e.g., anger toward perceived confederates). Some of the most delicate effects on other psychological states and processes are described by (Sloman 1987), (Weiner 1986), and (Bach 1992), (Frijda 1996) and include the arresting, interruption, and reordering of goals, as well as dramatic shift of focus of attention. (Shift in focus of attention may fall in the first category of unoverridable side-effects, or in the second with common impulses.) Such effects may even consist in changes to one’s various mental-state forming mechanisms, such as being more careful in the future of hastily judging wrongdoing or calibrating more carefully the strength of one’s feeling to the believed degree and likelihood of wrongdoing. Physiological effects (if any) are products of the feeling alone--it will not matter at whom, or over what, we are angry--while behavior, actions, and cognitive effects will be produced by the kind of feeling together with its cognitive object.

In this analysis, the very nature of anger, and what differentiates it from its first cousins rage and annoyance, are intended to be structural, and anger, as opposed to the anger-like rage and annoyance, crucially depends on relations to the producing belief about the morality of another person’s action. It is customary in 20th century article-length philosophical analysis to clarify a concept by using other concepts that are themselves often deeply complex and unexamined: knowledge, for example, in terms of "belief." By merely invoking the concepts of "belief," "person" and "moral judgment," I might seem to be proceeding likewise. I am aware of the lacuna, and pained by it, although I can gesture in the direction of how a concept such as "moral judgment" is itself to be clarified. This in turn will underscore the structural nature of all concepts, and the ways in which states of mind and their intentional objects are inextricably interwoven into a large, varied, but connected tapestry of mental life.

For a belief about a person’s action to be a "moral" judgment, we might guess that it involves some of these notions, and all of these features are ultimately amenable to a structural analysis. First, it shows a capacity and a tendency to recognize this type of action from others, to place it in a category (e.g., lying). Second, in order to be distinctively moral (and not just selfishly annoyed), it must indicate a "universal" tendency or mechanism to judge everyone, including oneself, by this standard: acts of this sort by any person are repellent. Third, the exact nature of this reaction is so constructed that it will involve or invoke emotions of guilt and regret if one notices oneself performing this action, and anger and resentment if one notices others doing it. Fourth, it will involve or invoke emotions of pride in oneself, and respect or admiration when one notices others avoiding its performance, particularly when the practical reasons for performing it are very great, that is, when temptation is great. The crucial notions, then, are a discrimination of this type of action, a universal, self-inclusive judgment, and then relationships with a variety of emotions that are distinctively associated with, and hence partly constitutive of, human moral judgment. Still more important, and deeper, is our precise conceptualization of the performer of this action as a person. As is well-known, in its full emotive and cognitive dimensions, this is a tremendously difficult and subtle notion. It is also one that is integrated into our most vital and significant experiences and states of mind. Roughly, we might guess that this is to have some developed conception and emotions about our own makeup (a reflective, full-bodied self-consciousness), and to believe that another creature is "likewise" able to feel, have thoughts and intentions, and so on. The richness and obscurity of this notion is obvious, but here impossible further to address.

We are now in a position to offer a general definition of an emotion. An emotion is a feeling that has been caused by certain beliefs, that is directed toward a primarily conceptual and not perceptual target (typically a person or intentionally alterable circumstance), and that typically produces some physiological, behavioral, or cognitive effects. We can compare this with (Sloman 1986 218): "...powerful motives respond to relevant beliefs by triggering mechanisms required by resource-limited intelligent systems" and (Averill and Nunley 1992 43) that emotions seem to include "passivity,... subjectivity,... and non rationality [in being] not considered rational or logical in the usual sense." In a masterful article (Frijda 1996), Frijda emphasizes the role of intensity and of seeing emotions involved in complicated ways in creating and sustaining socially consequent behavior (see also (Ortony et al. 1988, Ch. 4) and, for a purely social account, (Warner 1986)). Personal emotions are directed toward fellow persons, so conceived, or toward personified objects, and spawn intentions to aid, ignore, sustain, or impair their well-being. Impersonal (orthotic) emotions are directed toward states of affairs or non-personal entities conceived to be alterable, sustainable, or eliminable.

These emotive feelings have a strength or degree (I assume) although this is easily confused with the extent of the behavioral and physiological effects. In orthotic or "correct" cases of emotion, the intentional object will be of a certain type and this intentional object will be connected to the spawning beliefs in causal and conceptual ways. Thus in the case of anger, the proper object is the concept of an individual, conceived as a person, and this object is connected to the spawning belief that this very person (under the same description) intentionally performed some action. The various effects of an emotion will be highly variable, depending on the holder’s emotion-generating mechanisms, on other beliefs, goals, and emotions, individually or culturally ingrained habits of and constraints on emotional expression and action, and so on.

It is customary to describe an emotion, as well as a mental state such as a belief, as a disposition. Few authors are careful to describe whether the disposition is precisely the emotion (and if not, what else is required), or whether this disposition is constitutive of, or a conceptual part (or whole) of the state of mind itself, as opposed to being merely indicative of, or more distantly linked to, it. Specifically, the disposition associated in this way with an emotion or other mental state is often called a multiform disposition, and is generously multiform in recognizing that the disposition results in different behavior in different situations. Extreme such cases of the view are seen in the James-Lange theory of emotions, Gilbert Ryle’s account of mind and mental states, as well as in modern theories of mind, such as those motivated by Alan Turing’s "behavioral" criteria for intelligence.

A variant of the view that mental states "are" dispositions, is that a belief or state of mind is, or entails, a "commitment" or "conviction" even if it does not manifest itself in a first-order disposition. Namely, if I like my father but do not show this attitude in the situations we would expect some consonant behavior, or perhaps even seem to display an attitude of disrespect or dislike contrary to this supposed liking, then I may nevertheless show my attitude of liking my father by a sincere effort to reform my behavior, that is, by an awareness and active steps to correct, my missing first-order disposition. I may try to force myself to say nice things about my father, ask people to remind me when I lapse into displaying a lack of affection for my father, enter therapy to dislodge my behavioral block, and so on. Seen in this way, some commitments are still dispositions, but second-order ones: observable dispositions to change my primary dispositions and bring my behavior into accord with my "true" state of mind. Those who invoke such meta-dispositions generally require that such dispositions, unlike primary ones, must be "actively displayed" that is, not themselves overridden.

I believe this general method, of defining states of mind as consisting of, or even partially constituted by, dispositions (or meta-dispositions) is mistaken in four ways. First, it takes the general approach of saying that anger consists of anger-acting behaviors. This is unhelpful and ad hoc. It does not really explain events or states of mind, but instead postulates the existence of observable-event-producing mechanisms. This would be like resting with an account in physical chemistry that chemical properties consist in "bond-making properties," that is, in valences, without pursuing a further analysis of occult electrons, orbits, forces, quanta, and so on. Good science is not so cautious and parsimonious. Second, it is indicative of a positivist-behaviorist approach to the mind that is methodologically flawed. Namely, an overly close association of a mental state with a disposition or behavior is obsessively concerned with how we determine or observe some third person’s state of mind, rather than what it is. It confuses the criterion of determining or confirming a phenomenon (and only in third-person cases) with the phenomenon itself. Having anger is not the same as expressing or betraying anger. This is easy to see in the phenomenon of acting or pretending, which as we know from theater and everyday life, can be remarkably convincing. In fact, we are acquainted with anger most thoroughly and certainly through the first-person case, and proceed by imputing states of mind "like these" to others. Of all the phenomena in the world we form conceptions of, this alternative direct access that we have to states of mind is rare. We do not, for example, have first-person access to valences or electrons. It is this generalization--and the successes of the physical sciences--that encourages us to apply only this method to the theory of the mind as well. But I would argue that it is absurd to rely only on the manifestation of states of mind by others through our senses in order to give an account of them. This begins with the third-person instance as a base, or only, case and ignores our evident, and better (if mysterious), first-person experiences of them.

Third, no one ever really gets around to telling us exactly of what one of these dispositions consists. Anger presumably consists of a very large number of conditionals of the form "if S believes X, Y and Z, doesn’t feel more strongly about something else, and is in culture C and with normal neural connections... then S will scowl at P." It is unclear to me that we can describe, in a suitably precise way, wherein exactly such a disposition consists--nor can we, I believe, describe all the varying patterns of expression of anger for diverse individuals in diverse social circumstances (and cultures!) In short, we have not analyzed anger by merely saying that it is a "disposition." We have to give this disposition, or at least produce the method that would generate the disposition. Given the diversity of possible circumstances, and the possibility of ever higher-order overriding mental states, this seems problematic. A still more telling objection, observable in the first person case, is that it seems perfectly logically possible that I would have anger (or some other emotion) toward another individual and resolve never to express or betray it. Therefore, my having anger on this occasion cannot be identified as simply being a disposition, no matter how artfully or completely described: it is a disposition that will never appear.

My proposal is then not to focus on observed behavior, which varies from individual to individual and is radically relative to other interacting states of mind. Rather, a state of mind is distinctive in its manner of production (in its genetic nature), in the nature of its intentional target, and in the sort of "feeling" that connects these two. I unfashionably view produced behavior or observable physiological states as inessential and undifferentiating side-effects. I would call the general approach phenomeno-structuralist: states of mind consist in feelings that are related to other states of mind in certain ways. Admittedly, the structuralist component might be thought by others to be a kind of functionalism. But I am not wed to thinking--and do not think it is useful to think--of useful mental relations as primarily causal. Furthermore, mental states are sometimes distinctive just in ways that are structural but not obviously causal. For example, they can be distinctive in being directed at a person rather than a state of affairs, or depending on a moral judgment. Likewise, the notion of being "directed at" is a relation all right, but not obviously identical or reducible to causal relationships, and the difference between a person-type object of thought and a state of affairs, is a strong difference in metaphysical structure.

While I do not think there is a problem in identifying a particular state of mind as unexpressed or unobservable anger, I think there is a serious problem in supposing that an individual could be so constituted that he or she has anger but never expresses anger toward anything. The problem is not that such a person has no state of mind whatsoever, but rather in the appropriateness of calling it anger. I believe it is essential that what we call anger involves the contemplation of action that harms, in some mild or serious way, the person at whom one is angry. The unpleasantness of this feature can be overcome by conceptualizing this harm as punishment, justice, and so on. In most cases, this inclination will be overridden, and hence be unobservable; in some possible cases, it will lead to observable acts of retribution or "punishment."

Consequently I would like to extend this definition slightly. To have an emotional state identifiable as, for example, anger, then the correct causes, feeling, and intentional objects are necessary. But it must also be the case that the agent has the same kind of feeling in this instance toward the intentional object that in some other cases does produce observable behavior of a certain sort--namely a distinctively angry reaction. If no disposition to act in this particular case is observed or observable, this may be because of its being overridden by other mental states (e.g., discretion) or because of some defect in the "actional" mechanisms, such as a psychological defect or a severing of the facial nerves that allow one to scowl.

This is a perilous, but not desperate, situation. It requires the agent to have some capacity for generalization and recognition of kinds of feelings: this is itself a feeling that another feeling, such as one associated with a kind of anger, is present, and is distinct from the angry feeling itself. The possibility of recognizing such a primary feeling through a secondary feeling will be aided through the similarity of behavioral responses to many of the cases, as well as by the similarities among the causes and intentional objects. It may also be aided by a conceptualization of this particular state, such as through a name or description in a language, although this is not strictly necessary for the recognition of such feelings.

Thus while an indignant anger that provokes certain responses--especially consideration of causing some harm, conceptualized as "justice"--is necessary for this form of anger, the display of having considered such harm is not necessary. The feeling is "felt" to be like states of mind where punishment is considered.

Feelings. By a feeling I mean what C.S. Peirce called a "firstness," a primitive, basic mental state. The use in my or a similar sense, and an historical justification for it, can be seen (Langer 1973), (Gardiner et al. 1937), and (Aylwin 1985 1-2). It has no content, conceptualization and (what is the same as conceptualization for me), no links with other feelings, causes, or effects. A feeling is an "atom" of consciousness, having no parts, no inner structure. Considered purely as a feeling, without regard to what I will call their "external structure," their context, feelings are indiscernible, and we cannot distinguish one from another. That any such mental state exists is a theoretical hypothesis with little direct evidence in its favor, and is related to notions with notorious, convoluted histories in the philosophy of mind.

My idea of feelings has some connection to past proposals of sensations, qualia, sense-data, Humean impressions, the phenomenological "given," and other such primitive or foundational mental posits. My theory also explains the fact that we would be tempted to use a word, "feeling," to describe both the having of an emotion and the sensation of a pinprick. I would admit (unlike some sense-data theorists) that unconceptualized feelings are rare or nonexistent in adult mental life. Or rather, they cannot be isolated from their associations and conceptualization (they cannot be described or named, for example) and they tend to evolve quickly into linked and conceptualized "states of mind." Thus they are essentially incommunicable. Startle responses, pains and pleasures (before their location in the body is localized or their quality recognized), as well as sensations (especially from the mono-dimensional senses of sound and touch) come close to pointing toward what I am suggesting exist. One of the primary features of human mental life is, I believe, the propensity to associate and relate feelings to one another; consequently, to isolate them is both difficult and against our nature. Emotions and other "higher" mental states precisely consist in these associations of feelings, as do conceptual objects themselves. We encounter a barrier related to that in describing isolated feelings when we attempt to separate out a pure, isolated sensation ("redness now") from a perception (e.g. of a field or shade of redness or a specific red thing), that is, from some recognition and processing of what this sensation indicates, or how it is related to other sensations. These difficulties are as old as Berkeley and Kant. Such related or associated sensations are often called perceptions, I sense red and perceive a red truck.

Feelings play two important roles in my theory of states of mind. First, they return the analysis of mental states such as emotion to phenomenology, to the first-person acknowledgment of a species of consciousness. This is not unimportant, I believe, and returns the philosophy of mind to the study of the mental, rather than purely the mechanistic or computational, as if these latter analyses could ever produce wholly adequate understandings of the life of the mind, given that we ourselves as analysts have them. Second, they play an organizing role in the "structures" that specific states of mind become, namely, they are the pre-glued joints that hold together causes, intentional targets, behaviors, and so on. Finally, feelings, owing to their propensity in the human case to further organization and abstraction, allow for the full range of human mental life that may include cascading, higher-level consciousness that is typical of human mental life. A higher-order feeling becomes associated with structures of specific occurrences of anger. It causes in us what eventually seems to be a distinctive angry-feeling, a feeling that we feel anger, without details of precise cause, or contemplated effects and target.

It is also my view that concepts, and propositions, as well as perceptions, are themselves feelings: they are feelings associated with generally complex structures of other feelings. Thus although the details of my concept of a dog is quite complex, my contemplation of "dogs" (even without the simulated paradigmatic visual image) seems sensation-like. The exact nature of these mental structures, and how they come to be about the world, is another story, and one that is inessential for our purposes here. A full theory of mental content, one not so beholden to linguistic or communicative content, needs to be developed.

Emotions, together with a broad category of mental states I will call "wants," are distinctive in being essentially or strongly motivating. By this I do not mean to suggest that they immediately produce behavior and action: only goals and intentions do this. Rather, motivating states of minds produce goals and intentions, and these in turn produce behavior and action--frequently through habitual, rather than fully conscious mechanisms, and often in ways that are mediated by deliberative processes that may include weighing other emotions and wants, means-ends reasoning using beliefs about instrumentality, and prioritizing the resulting goals and intentions among already-present ones.

Among the motivating states of mind, emotions are loosely speaking "reactive" responses to an environment, in being caused by beliefs about particular other persons or particular states of affairs "exterior" to the person forming the emotion. Wants are broadly of two sorts: highly conceptualized, such as wanting a higher salary, happiness, or others’ respect, and less conceptualized, perceptually-targeted, wants, that we may call desires. Wants are adaptive rather than reactive: their causes lie primarily within the wanter’s mind, and are at most reactivated by perceived particular external objects and states of affairs. Desires are directed toward specific imagined or perceived, and less highly conceptualized intentional targets, such as drink (thirst), food (hunger) or another’s body (sexual desire). The intentional target of a desire, a perception, is, to use Hume’s language, an impression or type of impression, while the intentional target of a conceptualized want is an "idea." A perception is, in my account, a state of mind that is closely associated with, and thus as a perception proximately constituted of, sensory feelings. A highly conceptual object will be dominated by less direct links to sensations. The judgment that an object is "red" is more perceptual than the judgment that an object is a "house," since the latter will include many aspects of the object, as necessary conditions, that are not themselves sensory, such as that the object is an artifact or is an abode of human beings. Their distinction is to be made by the relations and associations of the feelings, intuitively by their "distance" along a graph of associations from basic feelings (such as sensations), and by certain parallels of this proposal with the Early Modern Theory of Ideas that is suggested by the Empiricists. A perception is a higher-order skeleton or pattern in a more detailed structure of feelings (e.g., sensations); it may eventually come to be habitually associated with a distinct feeling that then gives us the sense of immediacy and phenomenological primacy.

Primitive emotions, which I have called "reactive states of mind," such as the startle response, contrast with emotions in three ways. First, they have perceptual, even purely sensual, rather than conceptual, causes. Such a reactive state of mind never includes an awareness of the nature of the cause, and is thus necessarily "unreflective." Second, they have no or at most a perceptual (or even purely sensory) intentional object. Third, they are unmotivating, in the sense that they will not induce habitually trained behavior or action, even if they do produce reflexive and largely physiological responses. Additionally, they are typically momentary. In these latter respects, they are unlike conative wants, desires, which may also have similar origins.

Moods contrast with emotions most strongly in not having an intentional object: one has little or no idea at or about what one is sad, agitated, or happy. Their causes are typically conceptual or evaluative--things are not, or are, going well--but with little conceptual detail and often with respect, not to any specific object or state of affairs, but "in general." There is often little awareness of this cause, whereas proper and efficacious anger is typically reflective. (Moods may notoriously have purely physiological "causes" of a fundamentally different sort, such as drugs, light, or chemical imbalances of neurotransmitters.) Moods are not naturally, directly, and effectively motivating, in part because of their lack of an intentional object: on what should we act? They can serve as a general incentive to limit, or increase action in, for example, depression and agitation respectively. Moods may endure for long periods of time, while emotions are typically short-term or at most retriggered when their intentional object or causing judgment is again the focus of attention. I may be despondent for weeks, but I am not literally angry at Albert for days, but only when I think about him or about his despicable deeds.

It is the doxastic states of mind that are most crucially involved in my account. These are usually described as "beliefs," or to use the older term, judgments. Integral as beliefs are to most modern "representational" philosophies of mind, to say anything about them at all is difficult and dangerous. (At the same time, few authors have defined or characterized what it is to be a belief.) We can guess that a belief is an intentional state of mind, whose intentional target is a proposition, sentence, or state of affairs. This intentional target is a complex entity, made up by linking two or more conceptual objects--e.g., the object or noun and the propositional function or predicate. In this respect, it contrasts with emotions such as anger, or desire, which I have held are directed at single intentional objects, and is more like annoyance (which is however directed more at the "truth making" state of affairs than at the representation of a state of affairs, a proposition), conceptualized wants, and so on. A possible exception is belief "in" someone or something, which signals, I suspect, more an emotion-like attitude or commitment to an entity or value than a doxastic attitude proper. Some have suggested that, apart from the characteristic of the formal nature of its object, a belief must be produced by processes that are "aimed at" the truth. I do not think this can be made sense of within the context of the philosophy of mind--since truth, if it means anything like the traditional correspondence rather than coherence notion, includes reference to a non-mental "external" aspect of the world. We have no way of determining when a state of mind is, or is not, aimed at the truth. Rather, I believe that to count as a belief, it must be produced by mechanisms that are "usually" contradiction-rejecting and are "evidentially positive" in some traditional inductivist way. This addresses only the distinctive causal origins, or better: structural bases, of beliefs.

Beliefs are also distinctive in their use, that is, in what behaviors and other states of mind they cause. If there are means-ends beliefs, for example, then they must be utilized by schemes of practical reasoning: if we have a goal or intention, and are engaged in "planning," then we consider using such means in some of our means-ends beliefs, and sometimes do use them, if we are capable of action at all. There are thus the formal characteristic of beliefs (intentionally directed at "propositions"), their genetic characteristics (in being caused by certain mechanisms), as well as their pragmatic characteristics (their usability, and proper use, in practical reasoning and behavior or action). Additional characteristics may include the ability to be aware of, and thus change, one’s belief-forming mechanisms, an ability to form higher beliefs about our beliefs, an ability to manipulate the components of one’s belief contents in certain ways and to share components of beliefs’ contents with other states of mind (such as some emotions), an ability to attach feelings to beliefs that reflect their felt certainty, centrality, commitment, various other dimensions that are either emotional or entail further beliefs, and an ability to produce behavior or action that involves other beliefs and emotions, such as to "add" an inferred belief, or to "remove" an emotion if we find that the conception of the world that made us angry at Albert was after all false or unlikely. What counts as a "belief" thus involves recursive elements of its relation to other beliefs (and belief-forming mechanisms themselves), and integrated aspects of beliefs’ ability to modify other states of mind, such as intentions and emotions. This account necessarily skirts or fails to touch on important aspects of belief such as its phenomenal aspects, and diverse kinds of doxastic beliefs, such as presuppositions or assumptions, or the special status of belief about self, as well as how emotive states, such as hopes, or "improperly" produced propositional states, such as fictions or myths, may nevertheless pragmatically function as beliefs in some or all respects.

There are two interesting questions inevitably posed by analysis of emotions. First, is it possible that there is a small set of "natural" emotions common to all cultures and developed human responses to the world. A related question, often confused with the first, is whether there are perhaps basic kinds of emotion out of which more complex emotions are constituted or are such that all emotions are variations of these. Another question is that of whether there are states of mind that are partly constituted by emotions (or in some other ways structurally involve them) but may include in some sense other states of mind.

To the second question--whether emotions are parts of other states of mind--"romantic" and perhaps other forms of love come naturally to mind. Sloman (Sloman 1988 228-9) calls them "attitudes" that are ..."collections of beliefs, motives, motive generators, and comparators focused on some individual, object, or idea" and in works of Foucault and (Averill and Nunley 1992) we can begin to appreciate the full historical-sociological aspects of these not-so-primitive passionate states. I think it is probably mistaken to call them a "set" of such more basic states of mind, since these states of mind might themselves have complicated structure (in the shallow, causal analysis: in romantic love, lust might "cause" or color beliefs in the positive worth of their object and cause a cognitive motive of wanting the object, a person, to be happy.) With repetition, such large structural complexes of feeling may themselves phenomenally coalesce and come to be associated with a "feeling" of love--as with any complex of feelings. Such an observation might explain both the obvious complexity of love worthy of the name and its appearance to initiates as being phenomenally simple and direct.

The question of universal, natural, or atomic emotions--two or three different notions in my methodology--is an intimidating theme. We seem to have not yet had time in contemporary philosophical psychology to define and think precisely about emotions proper, and to evaluate competing theories of them, and the welter of our own emotional life, let alone to appreciate a lively and plausible literature about disparity of emotional life across cultures. Boldly to construct an atomic theory of emotions now would be a bit like a third-grader fresh from a brush with arithmetic contemplating the depths of standard and non-standard number theory.

One problem is obviously with armchair speculation about universal or natural emotions. Writers who are not deeply acquainted with cultures not their own--most of us, surely--will obviously privilege their own delineations in various ways that are hard to detect. (Syntactic universal, if there are any, are going to be much easier to see than emotional universals.) Even among Western European cultures we are surprised to encounter emotions demarcated differently, or even an emotion is one that appears not to be present in another.

A promising suggestion of John Myhill (Myhill 1995) is that most cultures have emotions and emotional vocabularies that have two components, a universal one, and a component or parameter that is peculiar to the beliefs and values of that culture. Thus I strongly suspect that what I have called "anger" is peculiar to those cultures that have notions of person, justice, responsibility, and will like our own (maybe even that have notions of harm and "belief" like our own). While I permit myself this neutral observation, I reserve the option that something like our culture’s delineation of these concepts is, in some respects, better (or worse) than those of other culture’s. Likewise, we can see components among notions such as shame, embarassment, sin, regret, and guilt--each of which seems to require slightly different aspects of a world-view, and thus might be present in different cultures to varying degrees.

The literature on basic emotions is both fairly extensive and, I believe, extremely speculative. Consequently, we must be careful about taking our speculations too seriously, intriguing though they may be. Myhill (Myhill 1997 117) proposes a matrix that generates general, universal emotions like:

Positive Negative

Past, directed to someone gratitude anger

Past, not directed happy sad

Anticipate future eager fear

Relate to object/being joy disgust

The emotions indicated in the intersection of these parameters are intended to be very general vague ones that will have much more specific content (and presumably feeling) in a given culture. Perhaps one could even distinctly feel variants of one of these "generalized" emotion within one culture. One of Myhill’s main points is that Hebrew words for anger and gratitude are keenly dependent upon what he calls a biblical worldview--roughly going along with, or against, what God wants or has ordained, and that the resulting retribution or harm will be wielded by God. Biblical "anger" is thus more a prediction that "you’ll be sorry." This comports perfectly well with my view that our notion of anger (as personal indignation) is strongly embedded in a liberal, Modern, Western, worldview.

I believe we can use the theory I have proposed to extend Myhill’s view to a more refined three-dimensional matrix induced by these three classes of parameters::

Source Target Suggested action

Past Another person Sustained/enhanced

Projected future Self Destroyed/harmed

Responsive entity not

a full person

Non-responsive entity

State of affairs

Non-targeted

"Anger" in my strict sense of indignation is thus based on a past event conceptualized as such, directed at a person or entity conceived of as a person, with an "impulse" to destroy or harm that person "for" a deed that person has committed. Cultures may also distinguish different notion of the past: Americans sometimes manifest a concept of "distantly past" temporal grievances that diminish or undermine emotion formation, while other cultures with ancestral hatreds do not do so--indeed the deed or behavior was not perpetrated on me, but by his ancestors on my ancestors a long time ago and is therefore especially grievous. Our culture has only traces of linguistically contrasting targets so finely as I have: some might feel, as I do, that it is bizarre and wrong to speak of being "angry"--and certainly righteously indignant--at an infant, dog, or car. Whether there are any coherent possible negative emotions that are distinctively directed toward these entities and states of affairs (perhaps even metaphysically or conceptually even more finely), and whether they are socially or individually good or useful are far more difficult questions.

What is important to remember in these various forms of "anger," and their possible cultural variations, is the huge number of ways in which a person or culture might regularly and distinctively conceive of them--how that emotion comes to be "structured." This will hold especially for how an individual regards some situation as unpleasant or an action or behavior as objectionable. But it will also hold for variations in the articulation and differentiation of self-hood (e.g., one’s own past actions), or otherness or relatedness of this target or source (e.g., one’s ancestors). A person’s behavior irritates me, their actions anger me, and their appearance disgusts me--all precisely and naturally enough expressed.

Like any theory of anything, the theory of emotion, and specifically of anger, should be measured by such factors as how clear it is, how useful it is in analyzing and articulating finer cases (e.g., annoyance and rage), how useful and plausible it is in diagnosing common pathologies (e.g., displacement of anger at one’s dog), and how useful it is in addressing and solving the major puzzles and problems associated with the phenomenon. Like emotions themselves, there has been little clear and first-rate thought about the "major problems and puzzles" associated with emotions. To put the matter clinically, what problem is there in the theory of emotions that is analogous to Russell’s paradox in set theory?

Perhaps somewhat surprisingly, I think some of the best thought about emotions and how they arise comes out of the philosophy of the arts. A sympathetic consideration of the value of emotions in art is of course as old as Aristotle, but has recently been a preoccupation among aestheticians such as Susanne Langer, Leonard Meyer, Kendall Walton, Peter Kivy and Jerrold Levinson. An especially interesting explanation of the connection among art, cognition, and emotion was suggested by Ayn Rand, but the full attractiveness of this account, or even its coherence as a theory, will likely await its careful exposition in (Kamhi and Torres, forthcoming). These puzzles about emotions are especially acute with respect to music, since non-programmatic absolute music without words seems capable of causing vivid, intense, emotion-like states. In fact, the puzzle is a double one, since music without words seems capable of creating these emotion-like states without presenting us with a representation of the object of the emotion. It can make us sad, without making us consider a sad state of affairs or have any intentional object (unlike film or literature). It also seems to induce in us what we might usually consider undesirable, "negative" emotion-like states that we nevertheless are perfectly willing to endure, and in fact seek.

It is my main contention that the fascinating work that has addressed this important topic has nevertheless been severely handicapped by not being able to distinguish emotions proper from other related states of mind (such as desire or mood) and has generally lacked a definition of emotion, and an anatomy or typology for emotions. Emotions have been addressed as if they were simple, well-understood phenomena. Related to the lack of a definition is the deeper problem of lacking an analysis of emotions and a theory of the relationships among states of mind something like the one I have offered. Consequently, most discussions of art and emotion have remained on the surface, appealing to intuitively-understood elements of our own experience--an often-failing method fraught with obvious difficulty.

Let us consider first the problem that makes us seem perverse and even masochistic in seeking "negative" emotions. This occurs especially in tragic or melancholic music, which is not just intellectually recognized, but sometimes deeply felt as tragic or melancholic. I think this complex psychological fact can be explained by several further, but plausible hypotheses, and by a careful consideration of the interaction among various "parts" of the emotional state. There is a fairly easy problem, and then a harder problem. The easy problem is that we do not act truly sad, at least not for a long period of time. In the model of effects of emotion, it is obvious that for action, other states of mind must be present: for example, to punish someone for a wrong deed, I must believe that there exists a person that I can punish. In the case of fictional figures, this is not possible: they are not present, or even existent, and I cannot "hurt" them. In the case of emotions aroused by art, we will thus have the physiological symptoms of emotion, but not the more conscious and considered actions and behavior, because of the failure of appropriate beliefs about their targets and means-ends. (Thus far, matters are fairly obvious and accounted for well by, for example, Kendall Walton’s analysis of "make believe.") Additionally, art works are frequently experienced in a sequestered "frame of mind" for relatively short durations--something perhaps like the elusive aesthetic experience. Eventually we emerge from a Berkeley cinema into the bright California sun, stumble home in our deeply moved state, gradually submerging ourselves again in the sundry goals, values and beliefs of our regular lives.

The deeper and more perverse aspect of emotions in art is the one formed by our seeking, even preferring, negative emotions, and how emotion-like states could even arise without our cognitive consideration of, for example, a sad situation. To explain this almost shocking fact, we need to go beyond what I have previously said about feelings and states of mind. I think it is plausible to believe, first, that we are by nature seekers of strong (and maybe many) feelings, and second that the categorization of a feeling as "negative" is not a visceral property of the feeling itself, but the result of a fairly secondary, intellectual process of recognizing its place in a context of other feelings and states of mind. While the feeling is a brute fact, a Peircean firstness or a qualia, its negativity is a more sophisticated act of consciously realizing its structure and place. I would further hypothesize that, because of its importance in biological evolution for altruism, we are more easily directed toward empathy with others’ "pain" than their joy. We might more easily soften or ameliorate pains, but joys are more delicate creations. Consequently, sad feelings at other than our own situation are both more easily produced, and not intrinsically unpleasant. They are feelings like any other, and it is very difficult to see how else we would explain a kind of joy or indulgence we sometimes have in nursing or cherishing the extreme grief we have at a loved one’s loss. In contemplating a great many evil and seemingly cruel events in human and our own personal history, the hypothesis that humans crave strong feelings of all sorts (and that negative or destructive ones are relatively easily produced) seems useful in explaining the world. This does not make human beings intrinsically wicked, merely amorally directed toward "feeling" and not just "pleasure" or "happiness" in its Panglossian philosophical interpretation, or toward power in its dark Nietzschean one. Seemingly the only exception to this generalization of the broad attractiveness of feelings is pain, which is for most people intrinsically repellent (and attractive even for deviants only when part of larger mental states).

This addresses the issue of the artistic indulgence of negative emotions and their frequency in artistic experience. There may be other explanations for some of the ways music engenders other emotions and feeling. There may also be social- and individual-prudential value in educating and training us to have relatively precise, differentiated, and practiced emotions that we are then able to summon forth at the right moment: the right emotion, and hence the right forceful response, in the right situation, to which good literature habitualizes us. But we must still explain how music manages to produce sadness, or a sad-like quasi-emotional state. I have only explained why we would want it and why the cultivation of negative emotions is not rarer.

Eventually through our emotional experience, we come to recognize certain common patterns of emotion, certain similar structures among different mental occurrences. That we "sense" or even "feel" these recurrent patterns would possibly be a better word choice saying that we "recognize" them. For the more common of these, especially if there is some social value in identifying them, or even an innate disposition toward this categorization, we will have words ("fear"), or be able to describe them through other means (what you’d feel if you were walking peacefully walking and suddenly came face to face with a large, growling, bear.) Others are a bit more problematic to capture in a word, to describe in a phrase, or are not always even precisely shared or so categorized by all individuals in all cultures--melancholy, or the distinctively nostalgic melancholy of Sehnsucht.

What we come to think of as an attribute of mental states is not precisely a property of the feeling itself, but of its relationship to other feelings, of its "external structure." And it is an emergent property that may only be sensed after we differentiate and reflect upon its components and through repetition and habit come to have this emotion as a predictable response to similar events, recognized or sensed as being similar. Music, we might guess, is capable of duplicating this structure, without having precisely the right components of usual or real sadness or empathy at another’s sadness. It does so because it, like real-life events, unfolds in time and has certain other structural similarities--a tempo, a progression of intensity or dynamic, and so on. (This formulation is almost entirely due to Langer’s little-discussed observation, which I believe is sublime.) There may be other features, which I think are possibly less important and culturally variable, such as associations of key (minor-major) with broad categorization of moods or emotion, of tempi, and so on. The mere perception of a musical work, and the sensing of these structural element is not just a cause of an emotion-like state, it duplicates key components of an emotion, and so is an emotion-like state. That is because the emotion (or at least a state of mind very similar to it) is identified precisely and only with structure, with the relations of feelings to each other. This would explain the vivacity of emotions we feel in music. We have the feeling we have come to attach to various contextualized anger-feelings, but without their context of precise causes, effects, and intentional object.

A great deal more remains to say about the structure of types of emotion, and how we sense them. Likewise, it is also important to recognize that adults, as when they fear the fictions in film, act (somewhat) differently and perhaps differentiate the real feeling of sadness or tragedy from one induced by an art work. This may because the actional and behavioral mechanisms are blocked by the absence of co-triggering necessary beliefs, by accompanying beliefs maintained throughout that "no one is really sad" or "I am sitting in my living room and nothing is really wrong," or because of more detailed structural differences between "real" emotions and the emotion-like states induced by music, even if we sense the similarity and for purposes of satisfying our craving for feelings, ignore the differences.

One difficult issue that separates modern philosophers of art is the view of what D. Matrevers called "cognitivists", as opposed to non-cognitivists. "Cognitivists" in Matrevers’ sense claim that emotions require or are partly constituted by beliefs. Since we do not "literally" have beliefs about fictions--we do not consistently act that way, for example--then cognitivists argue that it follows that in reacting to most art works, we do not have true emotions: we have quasi-emotions, or something to this effect. Although my own approach is broadly cognitivist--in seeing connections of emotions to our other concepts and belief-like states--I am, with Matrevers, an anti-cognitivist in his sense. I believe that the arguments of Matreversian cognitivists are ultimately unconvincing, and that they are succombing to a myth of the fully understood nature of belief, and perhaps especially, to an implicitly positivistic-behavioristic model of believing. I am reasonably certain that I feel sadness at Juliet’s death. I am less certain whether I "believe" or fail to "believe" that she is, at the end of Shakespeare’s play, dying. If my daughter asks me, as we are watching the end of the play, "Is Juliet dying?" or even "Do you believe Juliet is really dying?," I believe that the answer I would give is an unqualified: yes. So there are contexts in which I will properly say that I believe she is dying. About a philosopher’s literal sense of belief, I think I need to hear more before I will assent or dissent from saying that I "believe" Juliet is dying. The 20th century philosopher will quickly rush to argue with me: but don’t I realize that I am not acting as if I believed that someone was dying in front of me? Here the reasonable person should become wary. True, but I am certain that I probably have other beliefs that I do not consistently act upon--otherwise hypocrisy would not exist. I may feel I believe and even act on them in certain contexts, but not in others; I may believe episodically and momentarily. I don’t know. While certain patterns of behavior are symptoms or indicators of our beliefs, we perhaps should not rush to think that they are absolutely required for them or even that believing is partly constituted by these patterns of behavior. As Peirce astutely observes:

We believe the proposition we are ready to act upon. Full belief

is willingness to act upon the proposition in vital crises, opinion is

willingness to act upon it in relatively insignificant affairs.

Quoted in Nagel, The Last Word p. 127

Peirce, Reasoning and the Logic of Things, p. 112.

A full theory of belief-like states would then admit that there are some "strong" forms of belief that lack critical symptoms of behavior (hypocrisy), and that there are a range of belief-like states (including one’s more carefully distinguished than Peirce’s list) that are characterized by willingness to act in various circumstances: always, rarely, where there is great risk, and so on. Furthermore, this observable behavior is not itself the whole of what it is to characterize or define believing, but only one parameter or aspect of the usual way of speaking about belief--how it is connected to an individual’s actional mechanisms.

In conclusion, I am more certain that I really have emotions when I react to art than I am certain that I do not "believe" in some sense at that moment what some hold is only fictionally or make-believe true. To claim we have a clear and widely understood sense of "belief" that prevails on this point in yielding an answer of "No, I do not believe that"is slight-of-hand.

The account of emotion I have proposed is partly definitional. But is also introduces a theory of emotion and indeed, of all states of mind. I believe it is impossible to separate the two tasks. Likewise, the account is partly descriptive and partly revisionary. I take myself to be talking about what we usually call "emotions" in English, but have refined this term so that it excludes states of mind that some people might consider emotions (namely what I have called desires and moods, as well as quasi-doxastic states). And especially within my account of one emotion, I have appropriated for my own analytical purposes English words with less sharply delineated, or even different, meanings, such as "anger," "rage," and "annoyance." The account is also creative, or deviant, rather than purely conventional, in reaching outside of contemporary issues and vocabularies in the analytical philosophy of mind--to notions such as feelings, an historical appeal to the perception/conception distinction, a nonstandard account of conceptions, rarified ideas of structure, and rejecting so thoroughly a central role for dispositions and mental causation.

In its refusal to identify mental states with, or constituted from, observable behavior, it is anti-behavioristic. (This is not to say that observable behavior is not relevant to the correct judgment of when a given mental is present; but it does not consist of that behavior.) In its refusal simply to apply mechanistic and causal notions--and especially in a reluctance merely to import a naive conception of success in physical sciences through the application of physical "cause" to the mental--it is anti-functionalist. In the central role it gives to primitive conscious states--pure conscious events that I call "feelings"--it is phenomenologically realistic. In its emphasis on the relatedness of feelings to other feelings, and categorization of these by patterns of such relations, it is structuralist.

I have also interwoven neutrally describing emotional states with normative remarks about good--what I would prefer to describe as correct or orthotic forms of--emotions. I did so in part to show the use my theory of emotion might have for moral theory, in which I take emotions to have a very important role. As perhaps one of the most important aspects of moral psychology, emotions have this importance for three reasons. First, emotions are crucially motivational: they are, with wants, the states of mind that crucially lie on the series of events and entities that get us to do anything. Second, they often go wrong, in ways that destroy lives and disrupt the achievement of goals, and that traditional ethics and philosophy of mind have largely ignored. Third, in addition to their crucially motivating roles as responses to a particular world, they also have a fundamentally positive role that is often overlooked. Anger, for example, particularly as I have characterized it in its paradigm form of indignation, is a necessary impetus for seeking justice. Only through indignation can we ever be expected, either ourselves or through our governments, resolutely to seek the risky benefits of punishment. It is not for nothing that Aristotle focused, in his accounts of emotion in art and rhetoric, on the training of the proper forms of terror, pity and anger. It is vital in political life that precisely these emotions be "correctly" felt. Terror reinforces self-preservation and animates individual and the state’s defense. Pity produces compassion and a spirit of mutual aid. And anger produces a passion for justice. Like the delicately balanced system of counteracting forces in the solar system, such emotional forces, when carefully calibrated, hold a state together. Politically worthwhile emotional literature trains precisely these emotions, in the right way, and we thus have a ready defense against Plato’s charge of the uselessness, corrupting, or contemptible nature of fictional art, especially when it arouses emotions.

Finally, my attempt to focus on the emotions has strayed far beyond them, and instead wandered into characterizing other states of mind as well. This however was not just an idle digression, but part of my purpose in precisely situating emotions among their kindred states of mind. I believe we can understand a state of mind, and its role in human mental life, only be examining its complex relationships and interactions with other states of mind, and the complexly connected "feelings" that lie at their core.